Lectures 1-6 Pengukuran

Measurement Systems
Lecture 1- General Concepts
Hamid Ahmadian
School of Mechanical Engineering
Iran University of Science and Technology
ahmadain@iust.ac.ir
Types of Applications of
Measurement Instrumentation
• WHY STUDY MEASUREMENT SYSTEMS?
• CLASSIFICATION OF TYPES OF
MEASUREMENT APPLICATIONS
• COMPUTER-AIDED MACHINES AND

PROCESSES
WHY STUDY MEASUREMENT SYSTEMS?
 “Measurement system” includes all components in
a chain of hardware and software that leads from
the measured variable to processed data.
 Let’s introduce some basic ideas using the

automotive industry as an example.
 Modern automobile uses as many as 40 or 50

sensors (measuring devices) in implementing
various functions.
• In design stage, one must be aware of the
instruments available
▫ for the various measurements, and
▫ how they operate and interface with other parts of the
system.
▫ keep up with new sensor developments to allow
improvements in car design and operation.
• Lack of such sensor knowledge can severely restrict
the range of designs that one can conceive, thus
limiting improvements in overall car performance.
• Laboratory testing and the associated
measurement systems are a vital part of the
design process:
▫ If a new material is being considered, we may need to
run strength tests to develop data needed by the
design engineers.
▫ Or, a new or revised manufacturing process may
require statistical response surface experiments to find
the effects of process variables on performance and/or
cost.
▫ Finally, availability from suppliers of new components,
such as improved shock absorbers, may require
performance testing to decide whether their use is
warranted in the new design.
• As design and development proceed, prototype
subsystems and finally entire vehicles will be
produced.
• These are used as "test beds" to evaluate
performance and then feed back information to the
design/manufacturing teams.
• Once the design has been finalized, then
manufacture of the product in quantity can
commence,
▫ the manufacturing tools are controlled by a so-called
feedback mechanism
▫ some quality parameter of the part produced is
measured with appropriate sensors.
• The final product, a modern automobile, relies on a
multitude of sensors for its optimum operation:
▫ Speedometers tell us the vehicle's speed
▫ Tachometers display engine RPM
▫ Fuel gages keep track of the gas supply
▫ Temperature sensors warn of overheating
▫ GPS to locate the car and guide the driver
▫ Accelerometers signal air bags to deploy in case of a crash
▫ Brake-cylinder pressure and wheel-speed sensors control
the antilock braking system
▫ GyroChip to measure angular velocity to augment vehicle
stability during severe or emergency maneuvers.
CLASSIFICATION OF TYPES OF
MEASUREMENT APPLICATIONS
• Examples from any industry, can be classified
into only three major categories:
▫ Monitoring of processes and
operations
 keep track of some quantity
▫ Control of processes and
operations
▫ Experimental engineering
analysis.
• Every application of measurement can be put into
one of these three groups or some combination of
them.
General Concepts:
• Chapter 2: Generalized Configurations and
Functional Descriptions of Measuring
Instruments
▫ Functional Elements of an Instrument
▫ Active and Passive Transducers
▫ Analog and Digital Modes of Operation
▫ Null and Deflection Methods
▫ Input-Output Configuration of Instruments and
Measurement Systems
▫ Methods of Correction for Interfering and Modifying Inputs
• Chapter 3: Generalized Performance
Characteristics of Instruments
▫ Static Characteristics and Static Calibration
▫ Dynamic Characteristics
Measuring Devices
• Chapter4: Motion and Dimensional Measurement
• Chapter 5: Force, Torque, and Shaft Power
Measurement
• Chapter 6: Pressure and Sound Measurement
• Chapter 7: Flow Measurement
• Chapter 8: Temperature and Heat-Flux
Measurement
Fundamentals of Signal Analysis
• Time, and Frequency Domains
• The Time Domain
• The Frequency Domain
• Instrumentation for the Frequency Domain
Fundamentals of Signal Analysis
Understanding Dynamic Signal Analysis
 FFT Properties
 Sampling and Digitizing
 Aliasing
 Band Selectable Analysis
 Windowing
 Averaging
 Real Time Bandwidth
 Overlap Processing
Using Dynamic Signal Analyzers

References & Course Evaluation
Title: Measurement systems: application and
design
 By: Ernest O. Doebelin. 5th ed.
Publisher: McGraw-Hill series in mechanical
and industrial engineering
• Course Evaluation Scheme:
▫ Mid-Term 25% (chapters 2- 3)
▫ Lab Reports 35% (3 selected experiments)
▫ Final Exam 40%
Measurement Systems
Lecture 2- Generalized Configurations
and Functional Descriptions of Measuring
Instruments
Hamid Ahmadian
ahmadain@iust.ac.ir
Generalized Configurations and
Functional Descriptions of Measuring
Instruments
• Functional Elements of an Instrument
• Active and Passive Transducers
• Analog and Digital Modes of Operation
• Null and Deflection Methods
• Input-Output Configuration of Instruments
and Measurement Systems
▫ Methods of Correction for Interfering and
Modifying Inputs
Functional Elements of an Instrument
• In general without recourse to specific
physical hardware, one may describe both
the operation and the performance of
measuring instruments,
▫ The operation can be described in terms of the
functional elements of instrument systems,
 such scheme helps to understand the operation of
any new instrument with which one may come in
contact and to plan the design of a new instrument.
▫ The performance is defined in terms of the
static and dynamic performance
characteristics.
4

Simple Instrument Model
Key
functional
element
The Measurement
(Observed Output)
Mass of an
object Mechanical or Electrical
Weight
(Can be Manipulated in a
(Observable Physical Variable) Transmission System)
• A possible arrangement that includes all the
basic functions,
▫ It is a vehicle for presenting the concept of
functional elements, and not as a physical
schematic of a generalized instrument.
e.g. mechanical/electronic amplifier

Pressure gage
• The primary sensing element is

the piston, which also serves the
function of variable conversion
element
Pressure thermometer
• The liquid-filled bulb
acts as a primary
sensor and variable-
conversion element
▫ a temperature
change results in a
pressure buildup
within the bulb
Active and Passive Transducers
• One may group the instruments based on energy
considerations:
▫ a physical component may act as an active
transducer or a passive transducer.
• Passive transducer :A component whose output
energy is supplied entirely by its input signal
• An active transducer has an auxiliary source of
power which supplies a major part of the output
power while the input signal supplies only an
insignificant portion.
9
• Passive : Do not add energy but

may remove (thermocouple,...)
• Active : Add energy to the

measuring environment (radar,…)
• The electronic amplifier:
▫ the input-signal voltage, ei need
supply only a negligible amount of
power
▫ almost no current is drawn, owing to
negligible gate current and a high Rg
▫ the output element (the load
resistance RL) receives significant
current and voltage and thus power.
▫ power must be supplied by the battery
Ebb
▫ the input controls the output, but does
not actually supply the output power.
Analog and Digital Modes of Operation
• The majority of primary sensing elements are of
the analog type.
▫ For analog signals, the precise value of the
quantity (voltage, rotation angle, etc.) carrying the
information is significant.
• Digital signals are basically of a binary (on/off)
nature, and variations in numerical value are
associated with changes in the logical state
("true/false") of some combination of "switches.“
▫ the system is quite tolerant of spurious "noise"
voltages which might contaminate the information
signal.
Null and Deflection Methods
 Deflection Instrument
 For either static or dynamic measurements

 High dynamic response
 Energy drain from the measured …loading error
12
13
Null and Deflection Methods

 Null Instrument
- Key features
- Comparator for Iterative balancing operation
- Feedback to achieve balance
- Null deflection at parity
 High accuracy for small input values
 Low loading error
 Not suitable for high speed measurements
Measurement Systems
Lecture 3- Input-Output
Configuration of Instruments
and Measurement Systems
Hamid Ahmadian
ahmadain@iust.ac.ir
INPUT-OUTPUT CONFIGURATION OF
INSTRUMENTS AND MEASUREMENT
SYSTEMS
• A generalized configuration
containing the significant
input -output relationships
present in all measuring
apparatus,
▫ A scheme suggested by
Draper, McKay, and Lees
• Desired inputs : quantities
that the instrument is
specifically intended to
measure. • Modifying inputs are the
• Interfering inputs :quantities quantities that cause a change in
to which the instrument is the input-output relations for the
unintentionally sensitive. desired and interfering inputs
Examples: Interfering/Modifying inputs
The desired inputs p1

and p2 whose difference
causes the output x,
which can be read off If the manometer is not
the calibrated scale properly aligned with the
Measuring pressures under gravity vector, it give an
acceleration influence; an interfering output signal
error will be engendered (also a modifying input).
because of the interfering
acceleration input.
Modifying inputs: ambient temperature and gravitational force.
Both the desired and the interfering inputs may be altered by the modifying inputs.
Examples: Interfering/Modifying
inputs
• Interfering inputs:
▫ The 60-Hz magnetic field induces
voltages in the strain-gage circuit.
▫ The gage temperature causes a change
in gage resistance ;cause a voltage
output even if there is no strain.
▫ Temperature causes a differential
expansion of the gage which gives rise
to a strain.
• Modifying input:
▫ The gage factor is sensitive to
temperature
▫ The battery voltage Eb also changes
the proportionality factor
Methods of Correction for Interfering
and Modifying Inputs
• A number of methods for nullifying/reducing

the effects of spurious inputs are available:
▫ The method of inherent insensitivity
▫ The method of high-gain feedback
▫ The method of calculated output corrections
▫ The method of signal filtering
▫ The method of opposing inputs
The method of inherent insensitivity
• The elements of the instrument should inherently
be sensitive to only the desired inputs:
▫ Choosing gage material that exhibits an extremely
low temperature coefficient of resistance while
retaining its sensitivity to strain.
▫ In mechanical apparatus that must maintain
accurate dimensions in the face of ambient-
temperature changes, the use of a material of very
small temperature coefficient of expansion may be
helpful.
The method of high-gain feedback
• E.g.: Measuring a voltage ei by applying it to a

motor whose torque acts on a spring, causing a
displacement x0
Open-loop system
The method of high-gain feedback
• The output x0 is
measured by the
feedback device,
which produces a
voltage e0
proportional to x0
• We now require KFB

stay constant KAM to be very large
(unaffected by iM4) ("high-gain")
The method of calculated output
corrections
• Requires to measure or estimate the magnitudes
of the interfering and/or modifying inputs and
to know quantitatively how they affect the
output:
▫ In the manometer the effects of temperature on
both the calibrated scale's length and the density
of mercury may be quite accurately computed.
▫ The local gravitational acceleration is known for a
given elevation and latitude, so that this effect
may be corrected.
The method of signal filtering
Input filtering Output filtering

The method of input signal filtering
• Electromechanical devices for

navigation and control of
aircraft or missiles, • The interfering tilt-angle input
▫ The interfering vibration to the manometer may be
input may be filtered out by effectively filtered out by
use of suitable spring mounts. means of the gimbal-mounting
The method of input signal filtering
• The strain-gage circuit is

shielded from the interfering
60-Hz field
• The thermocouple reference

junction is shielded from
ambient temperature
fluctuations.
• Such an arrangement acts as a

filter for temperature or heat-
flow inputs
The method of output signal filtering
• The strains to be measured are mainly steady and never
vary more rapidly than 2 Hz.
• It is possible to insert a simple RC filter that will pass
the desired signals but almost completely block the 60-
Hz interference.
• The pressure gage modified by the insertion of a
flow restriction between the source of pressure
and the piston chamber,
▫ The pulsations in the air pressure may be
smoothed by the pneumatic filtering effect of the
flow restriction and associated volume.
• A "chopped" radiometer senses
the temperature Ts in terms of
the infrared radiant energy
emitted.
• The difficulty is that the
ambient temperature, as well
as Ts, affects Td.
• Interposing a rotating shutter
between the radiant source and
the detector, so that the desired
input is "chopped," or
modulated, at a known
frequency.
The method of opposing inputs
• Intentionally introducing into the instrument
interfering and/or modifying inputs that tend to cancel
the bad effects of the unavoidable spurious inputs.
• A millivoltmeter is basically a current-sensitive
device.
• However, as long as the total circuit resistance is
constant, its scale can be calibrated in voltage,
since voltage and current are proportional.
• This velocity increase
due to change in
streamline causes a
drop in static pressure.
• By properly choosing
distances d1 and d2 two
under/overpressure
effects can be made
exactly to cancel, giving
a true static-pressure
value at the tap.
• The mass flow rate of gas through
an orifice may be found by
measuring the pressure drop
across the orifice.
• Variations in gas temperature
and pressure yield different mass
flow rates for the same orifice
pressure drop.
• Opposing input is accomplished
by attaching the specially shaped
metering pin to a gas-filled
bellows, enabling the flow area to
be varied in just the right way.
• The action of the device is
that a vehicle rotation at
angular velocity θi causes a
proportional displacement
θo of the gimbal relative to
the case.
• When the temperature
increases, viscosity drops,
causing a loss of damping.
• Simultaneously, the nylon
cylinder expands,
narrowing the damping gap
and thus restoring the
damping to its proper
value.
Measurement Systems
Lecture 4- Generalized
Performance Characteristics
of Instruments
Hamid Ahmadian
ahmadain@iust.ac.ir
INTRODUCTION
• Study the performance of measuring instruments
and systems with regard to:
▫ how well they measure the desired inputs, and
▫ how thoroughly they reject the spurious inputs.
• The treatment of instrument performance
characteristics is broken down into the subareas of
▫ static characteristics, and
▫ dynamic characteristics.
• The overall performance of an instrument is then
judged by a semi-quantitative superposition of the
static and dynamic characteristics.
Static Characteristics and Static
Calibration
• Meaning of Static Calibration • Static Sensitivity
• Measured Value versus True • Computer-Aided Calibration
Value and Measurement: Multiple
• Some Basic Statistics Regression
• Least-Squares Calibration Curves • Linearity
• Calibration Accuracy versus • Threshold, Noise Floor,
Installed Accuracy Resolution, Hysteresis, and
• Combination of Component Dead Space
Errors in Overall System- • Scale Readability
Accuracy Calculations • Span
• Theory Validation by • Generalized Static Stiffness and
Experimental Testing Input Impedance: Loading
• Effect of Measurement Error on Effects
Quality Control Decisions in • Concluding Remarks on Static
Manufacturing Characteristics
Meaning of Static Calibration
• Static calibration :
▫ all inputs (desired, interfering, modifying) except
one are kept constant.
▫ the input-output relations is developed
• Superposition of these individual effects
describes the overall instrument static behavior.
• The calibration system should have a total

uncertainty four times better than the unit under
test.
Meaning of Static Calibration
• In performing a calibration, the following steps are
necessary:
▫ Examine the construction of the instrument, and
identify and list all the possible inputs.
▫ Decide, as best you can, which of the inputs will be
significant in the application for which the instrument
is to be calibrated.
▫ Procure apparatus that will allow you to vary all the
significant inputs over the ranges considered
necessary. Procure standards to measure each input.
▫ By holding some inputs constant, varying others, and
recording the output(s), develop the desired static
input-output relations.
Measured Value versus True Value
• The term "true value" refers to a reference value
that would be obtained if the quantity under
consideration were measured by an exemplar
method;
▫ a method agreed on by experts as being sufficiently
accurate for the purposes to which the data
ultimately will be put.
• If measurement process is repeated under
assumed identical conditions, we get a number of
readings from the instrument;
▫ the process must be in a state of statistical control
• Every instrument has an infinite number of
inputs;
• In a calibration procedure, certain inputs are
held "constant" within certain limits.
• These inputs contribute the largest components
to the overall error of the instrument.
• The remaining infinite number of inputs is left
uncontrolled, and it is hoped that each of these
individually contributes only a very small effect
• In the aggregate these effect on the instrument
output will be of a random nature.
• Effect of uncontrolled input on calibration
• A measurement process with good statistical
control generates a set of data exhibiting
random scatter.
Calibration was carried out Performing the calibration in a

without temperature control. temperature-controlled room
Some Basic Statistics
• Pressure-gage calibration data
• Histogram presentation:
•The area of a particular "bar" is numerically equal

to the probability that a specific reading will fall
in the associated interval.
•The area of the entire histogram must then be 1.0
• In limit with infinite number of readings, each
with an infinite number of significant digits, Z =
f(x) is called the probability density function.
Probability of reading The cumulative distribution

lying between a and b function
The normal or Gaussian PDF
Normal (Gaussian) Distribution
the population mean
Random variable
Non-dimensional variable
Clearance deviations
• Consider a shaft in a bearing DS=25.400 mm, and DB=25.451 mm. The
standard deviation of the shaft diameter is 0.008 mm, and the standard
deviation of the bearing diameter is 0.010 mm. For satisfactory operation
the difference in diameters (clearance) between the bearings must be
between 0.0381 mm and 0.0635 mm. What fraction of the final assemblies
will be rejected?
Mean Value
C  25.451  25.400  0.051 mm
STD
Sc  0.008  0.010  0.0128mm
2 2
0.0635  C 0.0381  C
 0.98,  1.01
Sc Sc
Clearance deviations
Considering C follows a Gaussian distribution 68.2% of products are accepted.
The remaining are rejected.
Qualitative test for conformity to the
Gaussian distribution
To plot Gaussian line, one must estimate

The effect of sample size increasing
• When real distribution is nearly Gaussian, if more

readings were taken, 99.7 percent would fall within 10.11
± 0.42 kPa
Display an empirical quantile-quantile
plot
Student's t Distribution
• Continuous, symmetrical distribution, used for
analysis of the variation of sample mean value for
experimental data with sample size less than 30.
▫ If the sample size is small (n < 30), the assumption
that the population standard deviation can be
represented by the sample standard deviation may
not be accurate.
▫ Due to the uncertainty in the standard deviation, for
the same confidence level, we would expect the
confidence interval to be wider.
• For sample sizes greater than 30, Student's t
approaches normal distribution.
• In case of small samples, a statistic called
Student's t is used: the sample mean
the sample size
the sample standard deviation
degrees of freedom
Gamma function
• Like the normal
distribution, these are
symmetric curves.
• As the number of samples
increases, the t-
distribution approaches
the normal distribution.
• For lesser values of v, the
distribution is broader
with a lower peak.
Confidence interval for the t-distribution
The probability that t lies in the

confidence interval, between -tα/2
and tα/2 :
For example, for a 95 %

confidence level,
α = 1 - 0.95 = 0.05
and α /2 = 0.025
Student's t as a Function of α and v
Example :
• A manufacturer would like to estimate the mean failure
time of its products with 95% confidence. Six systems are
tested to failure, and the following data (in hours of
functioning time) are obtained: 1250, 1320, 1542, 1464,
1275, and 1383. Estimate the population mean and the
95% confidence interval on the mean.
If we were to increase the confidence level, the estimated interval will also
expand, and vice versa.
Interval Estimation of the Population
Variance
• In many situations, the variability of the random
variable is as important as its mean value.
• The best estimate of the population variance, σ2,
is the sample variance, S2.
• …….
Measurement Systems
Lecture 5-Least-Squares
Calibration Curves
Hamid Ahmadian
ahmadain@iust.ac.ir
Least-Squares Calibration Curves
• In instrument calibration, the
true value is varied, in
increments, over some range,
causing the measured value also
to vary over a range.
• The procedure is merely to
cover the desired range in both
the increasing and the
decreasing directions.
• The average calibration curve
for an instrument generally is
taken as a straight line which
fits the scattered data points
best.
• An example of pressure gage
calibration.
1
1
1
⋮ ⋮ ⋮
1
,x
x=
1.0823
-0.8470
• The model parameters are derived from scattered
data; it would be useful to have some idea of their
possible variation:
• The standard deviation of q0 ,
▫ if qi were fixed and then repeated measurement of

qo would give scattered values,
• The standard deviations of m and b may be found
from:
• Assume sqowould be the same for any value of qi,
= 0.208 kPa, = 0.0140 and = 0.0830 kPa
• Assuming a Gaussian distribution and the 99.7
percent limits ( ± 3s),
m= 1.082 ± 0.042 ,b=-0.847 ± 0.249
• The least-squares line gives:
• The qi value computed in this way must have some

error limits put on it
• If we were measuring an unknown pressure
using this gage and got a reading of qo, our
estimate of the true pressure would be:
qo b
qi  3Sqi
m

2
1  o
q b  Sqo
Sqi  
N 2  m
qi   2
 m
• In case of this gage sqi, = 0.192 kPa and for qo=

4.32 kPa, the true pressure would be 4.78±0.58
kPa.
• Calibration allows decomposition of the total error
of a measurement process into two parts:
• The bias, also called the systematic error (since it is
the same for each reading and thus can be removed
by calibration), and
• The imprecision, also called the random error since
it is, in general, different for every reading and we
can only put bounds on it, but cannot remove it.
Uncertainty Calculation using t-
Distribution
• Improvements with two major features:
▫ A simple method considers a standard deviation
calculated from a small number of points to be as
accurate as one gotten from a large number of
points.
 Statistical theory (confidence intervals) allows us to
adjust the uncertainty to suit the number of points.
▫ The second improvement substitutes for our 3s
limits (99.7 percent level) a limit analogous to 2s
(95 percent level).
Distribution
• For Gaussian and near-
Gaussian distributions, 99.7
percent puts us well into the
"tails" of the distribution.
▫ it takes very large samples to
get reliable results for
probabilities
• Since most engineering
samples are relatively small,
quoting uncertainty bands at
the 95 percent level of
confidence is more realistic
Distribution t distribution values for uncertainty
calculation
• Mandel gives a ± 95 percent confidence
interval, defined by two hyperbolas on
either side of the least-squares line.
• The "vertical" location of the two
hyperbolas as a function of qi is N-2
computed from:
Distribution
• For a read out (q0), we draw a horizontal line
through that value.
• This line intersects the two hyperbolas, and the
qi values at these two intersections define the
ends of a 95 percent confidence interval for the
true value.
Distribution
• Visually, the two "hyperbolas" seem to be
straight lines, but inspection of the tabular
results shows that Δq0 does vary with qi
▫ the largest value being at the left and right ends of
the curves and the smallest being at the center.
Distribution
• A major assumption of the analysis is that the statistical
variability of the measurements is the same over the entire
calibration range.
• It is good practice to also make a plot of the residuals versus qi;
▫ Such a graph can show whether this assumption is reasonable.
• For our example, shows no obvious trend in the size of the
residuals; the variability seems to be about the same over the
whole range.
Pressure gage imprecision
• The average calibration curve for a pressure gage
is q0  1.0823qi  0.8470, Sq0  0.208kPa
• For qo= 4.320 kPa, what is the best estimate of
the true pressure?
qo  0.8470
qi   4.691 kPa
1.0823
• Calculate the imprecision in the estimation of
input pressure with the 95% confidence interval.
0.208
qi  2 Sqi  4.691  2  4.691  0.384 kPa
1.0823
  4.307  5.075
qo= 4.320
4.307-5.075
qo

• A 95 percent confidence interval can be defined by two hyperbolas on either
side of the least-squares line where the "vertical" location of the two
hyperbolas as a function of qi is computed from:
• Calculate qo for the pressure gage when the input pressure is 5 kPa. Note:
The experiments is repeated twice (n=2).
1 1
N  22, qo   2.086  0.208    0.  0.32 kPa
2 22
• At which input pressure(s) qo is maximized?
@ 0. & 10 kPa
Measurement Systems
Lecture 6- Calibration Accuracy/
Overall System-Accuracy Calculations
Hamid Ahmadian
ahmadain@iust.ac.ir
Calibration Accuracy versus Installed
Accuracy
• It was stated that calibration removes the bias
portion of the error,
• This is true only for the conditions under which the
calibration was performed:
▫ This means that the measurement error (bias and
imprecision) must be re-evaluated, taking into
account, as best possible, the deviation of the
measurement conditions from the calibration
conditions.
▫ This re-evaluation is usually not as straightforward as
the calibration was because the measurement
environment is rarely as controlled as a standards
laboratory calibration.
Differences between calibration and
measurement situations
• A simple spring-type force measuring scale could
easily be calibrated with standard masses,
▫ find a best-fit line and uncertainty, and remove any
scale bias present.
• If the temperatures at calibration/measurement are
different, the scale will exhibit an uncorrected bias
with two sources:
▫ thermal expansion (which shifts the zero point) and
▫ temperature sensitivity of the spring's elastic
modulus (which changes the spring stiffness).
• Other possible effects include angular misalignment
of the unknown force with the scale's sensitive axis.
Accuracy
• One aspect of the measurement situation is that the bias
portion of the error is now not zero. (Recall that we earlier
said that calibration removes the bias.)
• Biases are classified into five different types:
▫ Large known biases (eliminated by calibration).
▫ Large unknown biases(not correctable; usually come from
human errors in data processing, incorrect installation and/or
handling of instrumentation, and unexpected environmental
disturbances).
 In a well-controlled measurement process, the assumption is that
there are no large unknown biases.
▫ Small known biases (may or may not be corrected, depending
on the correction difficulty and their magnitude.).
▫ Small unknown biases with unknown algebraic sign.
▫ Small unknown biases with known algebraic sign.
Accuracy
• Small, unknown biases remain as a contribution to
the measurement error.
• The bias in the measurement situation (as contrasted
with calibration) is treated as a random effect rather
than as systematic
▫ Bias limit: It is defined as the range of values within
which we feel that the actual bias will be found 95
percent of the time.
• Using this scheme, the "total error”/uncertainty in
the measurement is the sum of the bias limit and the
imprecision,
Accuracy
• To compute uncertainty in force
measuring scale , we need to estimate
the temperature, misalignment, and
any other effects felt to be significant.
• Note that:
▫ we do not measure the temperature
and misalignment and then correct for
these effects, rather
▫ we estimate some limits on how large
we think these effects might be and
then add this to the uncertainty.
Accuracy
• A displacement-measuring dial indicator
as part of an experiment to find the beam's
spring constant, F/δ.
• A bias error will be introduced because the
indicator spring force acts against F,
▫ Causing the measured deflection to be less
than it should be.
▫ If F is always downward, this bias error
would be treated as having an unknown
magnitude, but a known sign.
▫ The deflection is always measured too low.
• If we estimate an upper limit for its
magnitude, this bias would give an
unsymmetrical uncertainty; for example,
-0.003 in. to +0.001 in.
Accuracy
• Thermocouples are calibrated in an
accurately controlled and measured
temperature environment.
▫ The wires are immersed in a liquid-filled
well whose temperature is uniform at Thot
over a long distance to prevent conduction
heat transfer along the wires, which would
cause the sensing tip to read too low.
• When used to measure the temperature
of a hot gas, the wires are in contact with
a cool duct wall, conduction is now not
negligible, and the sensing tip will read
low.
Accuracy
• If such an error causes unacceptable
uncertainty, we may measure the wall
temperature, estimate the needed heat
transfer parameter, and compute a
correction.
• This correction will improve the uncertainty,
but not eliminate it, since the correction will
itself be uncertain, which uncertainty we
will have to estimate and include.
• For example, if our reading is 357°C and,
▫ the correction is +8°C, the nominal value is
365°C.
▫ If the uncertainty in the correction is±2°C,
and the uncertainty due to other sources was
±5°C, the temperature would be quoted as
365±7°C.
Accuracy
• In situ calibration, if possible, would in many cases be preferred,
▫ the calibration numerical results would include all the effects
contributing to uncertainty
▫ not require separate judgments and estimates based on experience
rather than actual measured data.
• In a similar spirit, we should also consider the end-to-end
calibration.
▫ rather than calibrating separately each link (sensor, amplifier, filter,
recorder, etc.) in our measurement chain and then combining the
individual uncertainties mathematically, we apply a standard to only
the sensor input and record only the final output.
▫ Advantage :all interactions among the links are automatically taken
into account and the procedure may be considerably quicker.
▫ Disadvantage: we do not see which components are contributing the
most to the total uncertainty.
▫ Even when we do perform the individual calibrations, a final end-to-
end study may be desirable.
Combination of Component Errors in
• A measurement system is often made up of a chain
of components, each of which is subject to
individual (known) inaccuracy.
• How is the overall inaccuracy computed?

• If the Δx's are now considered to be the
uncertainties uxi in each measured value xi, then
the corresponding uncertainty Uy in y is
• Think of the partial derivative as the sensitivity
of y to changes in the particular x.
▫ when a partial derivative has a large numerical

value, y is very sensitive to that particular x.
• The partial derivatives are numerically evaluated
at the operating point,
▫ they are constants (not functions), so this
equation defines y as a linear function of the x's,
even though the original function (f) may be
nonlinear.
• This relation is called the root-sum-square (rss)
formula,
• We have not here proven its validity, but this

rests on the fact that the standard deviation of
any linear function of Gaussian independent
variables is given by the square root of the sum
of squares of the individual standard deviations.
• It is an approximate result because y is not really
a linear function of the x's; it is close to linear
only for small changes.
Tuning Manometers Uncertainty
• Manometers pressure-measuring devices determine a pressure by measuring the
height of a column of fluid (P=ρgh). We would like to achieve an accuracy of 0.1 % of
the maximum reading, 10 kPa. This is to be done by using a type of manometer called
a well manometer, which has an uncertainty of 1/10 mm in reading the scale.
Estimate the uncertainty that can be tolerated in the density of a gage fluid, which
has a nominal value of 2500 kg/m3. It is assumed that the value of g is known to a
much higher degree of accuracy than the rest of the parameters.
P 10000
 P   P   P  h   0.408m 
 P         g     h  ,  g 2500  9.81
    g   h  2
 1      0.   1 
2 2 2
 P    h g     h   g    g   h  , 
 1000
   

    9.81
  
  4080 
P      g    h  
     .  0.00097  0.1%
P     g   h  
Tuning Manometers Uncertainty
• Calculus of variations:
Ln P  Ln   Ln g  Ln h,
P      g    h 
    ,
P     g   h 
2 2
  P       g    h 
2 2
        .
 P      g   h 
EXAMPLE: Uncertainty Calculation
• Consider a load transducer made by attaching a
strain gage at the root of a cantilever beam.
• A tip force, P, will produce a bending strain εx.
• Assume the electrical output of the strain gage is
e=Kεxe0 where e0 is the excitation voltage and K
is the calibration factor to be determined.
Uncertainty Calculation
1/2
 K   K
2 2
  K   K
2
  K
2
  K
2
  K
2
 
2
 K   e    e0    b    h   E   P   L  
 e   e 0   b   h   E   P   L  
1/2
 K   e    e 0    b    h    E    P    L  
2 2 2 2 2 2 2
        2        
K  e   e 0   b   h   E   P   L  
• Assume the following uncertainties:
Note: for 95% confidence interval (1.96 σ);

this affects only the precision error which
is purely random.
• Total uncertainty in the measurement of K is then:
K 2 1/2
 U bias  U precision 
2
K
2 1/2
  2.65  3.94 
2
%  4.75%

Lectures 1-6 Pengukuran

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lectures 1-6 Pengukuran

Uploaded by

Copyright:

Available Formats

Measurement Systems

Lecture 1- General Concepts

• WHY STUDY MEASUREMENT SYSTEMS?

• COMPUTER-AIDED MACHINES AND

 Let’s introduce some basic ideas using the

 Modern automobile uses as many as 40 or 50

Using Dynamic Signal Analyzers

Functional Elements of an Instrument

e.g. mechanical/electronic amplifier

• The primary sensing element is

Active and Passive Transducers

• Passive : Do not add energy but

• Active : Add energy to the

 For either static or dynamic measurements

Null and Deflection Methods

The desired inputs p1

• A number of methods for nullifying/reducing

• E.g.: Measuring a voltage ei by applying it to a

• We now require KFB

Input filtering Output filtering

• Electromechanical devices for

• The strain-gage circuit is

• The thermocouple reference

• Such an arrangement acts as a

• The calibration system should have a total

Calibration was carried out Performing the calibration in a

•The area of a particular "bar" is numerically equal

Probability of reading The cumulative distribution

To plot Gaussian line, one must estimate

• When real distribution is nearly Gaussian, if more

the sample size

the sample standard deviation

The probability that t lies in the

For example, for a 95 %

▫ if qi were fixed and then repeated measurement of

• The qi value computed in this way must have some

• In case of this gage sqi, = 0.192 kPa and for qo=

Pressure gage imprecision

• How is the overall inaccuracy computed?

▫ when a partial derivative has a large numerical

• We have not here proven its validity, but this

• Assume the following uncertainties:

Note: for 95% confidence interval (1.96 σ);

• Total uncertainty in the measurement of K is then:

You might also like